Golemcore Bot, from Alexk Dev, functions as a self-hosted agent runtime and an Agent Platform for AI-Native Companies. It is distributed primarily as a Docker image, ghcr.io/alexk-dev/golemcore-bot:latest, and can also be installed using a curl | bash script that employs docker compose. This platform integrates with various systems, including channel-connected agents, skills, plugins, and Model Context Protocol (MCP) servers. It supports major Large Language Model (LLM) providers like OpenAI, Anthropic, and Google AI, and can optionally connect to GolemCore Hive for fleet orchestration.

What does Golemcore Bot connect to?

Golemcore Bot connects to a range of external systems and services. It integrates with major LLM providers such as OpenAI, Anthropic, Google AI, and custom endpoints, enabling diverse AI capabilities. The bot supports connections to Model Context Protocol (MCP) servers and utilizes a plugin ecosystem for extended functionality. For fleet orchestration, it can optionally connect to GolemCore Hive. Communication with MCP tool servers uses stdio, while its dashboard chat interface operates via a live websocket. It also supports HTTP for incoming webhooks from systems like CI, alerts, or GitHub.

Core features and tools

Golemcore Bot provides capabilities through built-in tools, plugin tools, and MCP tools. A direct callable tool within the bot is set_tier for model routing. Plugin-provided tools include Brave Search, GolemCore Brain wiki tools, Playwright-backed browser automation, Browserless smart scrape, ElevenLabs-backed STT/TTS, Open-Meteo weather, and Email (IMAP/SMTP) tools. MCP servers, such as gulemui/gui-mcp, expose tools like json_generate_from_schema, json_generate_from_openapi, json_get_widget_spec, get_concept, json_validate_form_definition, dx_list_factories, dx_get_spec, and dx_check_code.

Who is it for?

Golemcore Bot targets AI-native companies and developers or operators managing agent runtimes. It is designed for those who need to run channel-connected agents, extend runtime functionality through skills and plugins, and configure model routing. The platform supports integrating AI agents into existing workflows, such as connecting to GitHub or monitoring CI. Its architecture caters to environments requiring custom tool integration and flexible LLM provider connections. Skip this if your organization does not require an agent platform for AI-native operations.

How to get started

The server authenticates using a single admin account. The BOT_DASHBOARD_ADMIN_PASSWORD environment variable can pre-set the password.

Use cases

• Running channel-connected agents for chat within the dashboard, Telegram, or via webhooks, with full inspection of execution, memory, and trace output.
• Extending the runtime by adding skills, installing plugins, connecting MCP servers, and configuring model routing tiers to expand capabilities.
• Connecting to GitHub via MCP for scheduled code reviews or monitoring CI systems with webhooks for automated task management.

Golemcore Bot is an adaptable agent runtime for AI-native companies, offering extensive integration with LLM providers, plugins, and MCP servers. Its self-hosted nature provides control over agent operations and data. The platform supports diverse communication methods and tool extensibility, making it suitable for complex AI workflows.